DeepLensNet: Deep Learning Automated Diagnosis and Quantitative Classification of Cataract Type and Severity.

PURPOSE: To develop deep learning models to perform automated diagnosis and quantitative classification of age-related cataract from anterior segment photographs. DESIGN: DeepLensNet was trained by applying deep learning models to the Age-Related Eye Disease Study (AREDS) dataset. PARTICIPANTS: A total of 18 999 photographs (6333 triplets) from longitudinal follow-up of 1137 eyes (576 AREDS participants). METHODS: Deep learning models were trained to detect and quantify nuclear sclerosis (NS; scale 0.9-7.1) from 45-degree slit-lamp photographs and cortical lens opacity (CLO; scale 0%-100%) and posterior subcapsular cataract (PSC; scale 0%-100%) from retroillumination photographs. DeepLensNet performance was compared with that of 14 ophthalmologists and 24 medical students. MAIN OUTCOME MEASURES: Mean squared error (MSE). RESULTS: On the full test set, mean MSE for DeepLensNet was 0.23 (standard deviation [SD], 0.01) for NS, 13.1 (SD, 1.6) for CLO, and 16.6 (SD, 2.4) for PSC. On a subset of the test set (substantially enriched for positive cases of CLO and PSC), for NS, mean MSE for DeepLensNet was 0.23 (SD, 0.02), compared with 0.98 (SD, 0.24; P = 0.000001) for the ophthalmologists and 1.24 (SD, 0.34; P = 0.000005) for the medical students. For CLO, mean MSE was 53.5 (SD, 14.8), compared with 134.9 (SD, 89.9; P = 0.003) for the ophthalmologists and 433.6 (SD, 962.1; P = 0.0007) for the medical students. For PSC, mean MSE was 171.9 (SD, 38.9), compared with 176.8 (SD, 98.0; P = 0.67) for the ophthalmologists and 398.2 (SD, 645.4; P = 0.18) for the medical students. In external validation on the Singapore Malay Eye Study (sampled to reflect the cataract severity distribution in AREDS), the MSE for DeepSeeNet was 1.27 for NS and 25.5 for PSC. CONCLUSIONS: DeepLensNet performed automated and quantitative classification of cataract severity for all 3 types of age-related cataract. For the 2 most common types (NS and CLO), the accuracy was significantly superior to that of ophthalmologists; for the least common type (PSC), it was similar. DeepLensNet may have wide potential applications in both clinical and research domains. In the future, such approaches may increase the accessibility of cataract assessment globally. The code and models are available at https://github.com/ncbi/deeplensnet.

Surgical simulator training reduces operative times in resident surgeons learning phacoemulsification cataract surgery.

PURPOSE: To compare the surgical duration for routine phacoemulsification surgeries in residents with and without virtual simulator training. METHODS: Retrospective cohort study of operative times of routine phacoemulsification cataract surgeries performed by 29 different third-year residents rotating at one academic institution. One group underwent mandatory virtual cataract surgery simulator training (SIM) in their second year of residency before starting surgeries while the other group did not undergo any simulator training (NOSIM). Outcomes measured were comparative surgical times and vitreous loss rates between groups in their third year of residency. RESULTS: 722 surgeries were included. Surgeries in the SIM group were on average 6.7 min (min) shorter compared to the NOSIM group (P = 0.0001). Although both groups required less time for surgery over the course of the academic year, regression analysis showed that NOSIM group residents overall required 17% longer time for an uncomplicated clear corneal phacoemulsification surgery (incidence rate ratio 1.17; p = 0.0001). In the final month of their residency residents in the SIM group (32.2 ± 3 min) were 9 min faster than NOSIM peers (41.2 ± 3 min mean ± SE; p = 0.02). Vitreous loss rates were 1.4% in the SIM group and 3.6% in the NOSIM group (p = 0.06). CONCLUSION AND IMPORTANCE: Early and continuous implementation of mandatory virtual simulator surgical training before starting intraocular surgeries significantly decreases operative times in third year residents learning phacoemulsification compared to non-simulator trained peers.

Eleven Years of Cataract Surgery in Veterans Without Pre-existing Ocular Comorbidities.

INTRODUCTION: In 2017, over 75,000 cataract surgeries were performed within the Veterans Health Administration System (VHA). Previous reports of outcomes of cataract surgery in veterans include patients with pre-existing ocular disease, which can affect vision. To exclude the confounding factor of pre-existing ocular comorbidities, we investigated the long-term visual outcomes and complications associated with small incision cataract surgery performed on veterans without any pre-existing eye disease. MATERIALS AND METHODS: Institutional Review Board approved cohort study with detailed retrospective chart review of all phacoemulsification (small incision) cataract surgeries performed at the Veterans Affairs Medical Center in Washington D.C. over 11 years, including all pre-and postoperative visits until postoperative month 12. RESULTS: A total of 1,513 consecutive surgical cases without any pre-existing ocular disease except the cataract were included. Vision improved significantly after cataract surgery compared to the preoperative best-corrected visual acuity (BCVA) (p = 0.0001) and remained stable over the first intra- and postoperative year. All eyes without complicated surgery and 99.1% of eyes with complications achieved 20/40 or better final vision postoperatively. The most common intra-and postoperative complications were vitreous loss (3.1%) and cystoid macular edema (CME; 1.4%). Patients with complications achieved final mean BCVA of 0.04 (20/22, vitreous loss) and 0.06 (20/23, CME) mean logMAR (Snellen). CONCLUSION: Analysis of 11 years of small incision cataract surgery in eyes without pre-existing ocular disease within the VHA showed significant improvement in vision and stability 12 months after uncomplicated and complicated surgery in veterans.

Chronic visual dysfunction after blast-induced mild traumatic brain injury.

The purpose of this study was to investigate the long-term visual dysfunction in patients after blast-induced mild traumatic brain injury (mbTBI) using a retrospective case series of 31 patients with mbTBI (>12 mo prior) without eye injuries. Time since mbTBI was 50.5 +/- 19.8 mo. Age at the time of injury was 30.0 +/- 8.3 yr. Mean corrected visual acuity was 20/20. Of the patients, 71% (n = 22) experienced loss of consciousness; 68% (n = 15) of patients in this subgroup were dismounted during the blast injury. Overall, 68% (n = 21) of patients had visual complaints. The most common complaints were photophobia (55%) and difficulty with reading (32%). Of all patients, 25% were diagnosed with convergence insufficiency and 23% had accommodative insufficiency. Patients with more than one mbTBI had a higher rate of visual complaints (87.5%). Asymptomatic patients had a significantly longer time (62.5 +/- 6.2 mo) since the mbTBI than symptomatic patients (42.0 +/- 16.4 mo, p < 0.004). Long-term visual dysfunction after mbTBI is common even years after injury despite excellent distance visual acuity and is more frequent if more than one incidence of mbTBI occurred. We recommend obtaining a careful medical history, evaluation of symptoms, and binocular vision assessment during routine eye examinations in this prepresbyopic patient population.